Microsoft: have deployed a cutting-edge AI model for U.S. intelligence agencies to analyse top-secret information. This GPT-4-based model was unveiled at the 2024 SCSP AI Expo for National Competitiveness in Washington, D.C., and it’s a game-changer. What’s special about this model? It’s completely cut off from the internet, making it super secure for handling classified data.
William Chappell, Microsoft’s chief technology officer for strategic missions and technology, explained, “This is the first time we’ve ever had an isolated version – when isolated means it’s not connected to the internet – and it’s on a special network that’s only accessible by the U.S. government.” The goal is to ensure that the model doesn’t accidentally reveal any sensitive information it learns from the questions asked.
Intelligence communities have long been searching for an AI model like this, but security has always been a concern. Chappell assures that this new model is secure, thanks to being in an “air-gapped” cloud environment that keeps it isolated from the internet. Unlike most AI models that learn from uploaded files, this spy model won’t do that. This way, the government can keep the model “clean” and prevent any secret info from being absorbed.
This super-smart model operates entirely separate from the internet, making it a first of its kind. Last year, the Central Intelligence Agency (CIA) introduced its own ChatGPT-style AI tool, but it was only used for sorting through public info—nothing classified. Other government agencies, like the Pennsylvania Office of Administration, use generative AI, but with non-classified data.
Microsoft has been hard at work on this top-secret project for the past 18 months, even revamping an AI supercomputer in Iowa to make it happen. The model has only been live for less than a week and still needs to undergo testing and accreditation by the intelligence community.
This launch couldn’t come at a more critical time for Microsoft’s cybersecurity rep. The Department of Homeland Security recently slammed Microsoft’s security practices, blaming them for exposing the emails of high-ranking government officials. CEO Satya Nadella has since made security the company’s “top priority.”
But Microsoft isn’t shying away from developing AI tools for government agencies. Just last month, it was revealed that Microsoft Azure’s version of DALL-E was pitched as a battlefield tool for the U.S. Department of Defense. While Microsoft has a long history of working with militaries and governments, their foray into AI is relatively new.
